00 vote positif00 vote négatif

8 vues15 pagesarticulo de modelación

Feb 10, 2015

© © All Rights Reserved

PDF, TXT ou lisez en ligne sur Scribd

articulo de modelación

© All Rights Reserved

8 vues

00 vote positif00 vote négatif

articulo de modelación

© All Rights Reserved

Vous êtes sur la page 1sur 15

the point estimation of landll gas surface

emissions using the ux chamber technique

M.D.N. Perera, J.P.A. Hettiaratchi, and G. Achari

Abstract: The accurate determination of gas emissions from sanitary landfills is necessary to assess the contribution of

landfill gas emissions to global warming and to develop optimum techniques to control emissions from individual landfills.

The most popular method available for point measurement of landfill gas, the closed or static flux chamber technique,

has inaccuracies associated with the technique itself. A new mathematical modeling approach is presented to estimate

the errors caused by changes to the flow pattern of gases within the soil cover when a closed flux chamber is placed

on top of the cover to measure gas emissions. The proposed two-dimensional numerical model considers both advective

and dispersive processes of gas migration. The model was calibrated and verified using data from controlled laboratory

experiments. The model was able to predict the fluxes with 95% accuracy. Based on laboratory experiments and model

predictions, our analysis showed that a closed flux chamber technique would underestimate actual gas flux by as much

as 65% depending on the configuration of the flux chamber, the actual flux rate, and the characteristics of the soil in the

landfill cover.

Key words: landfill gas, methane, global warming, mathematical modeling, landfill cover, simulation, environmental

monitoring, flux chamber.

Rsum : Il est ncessaire de dterminer prcisment les missions gazeuses provenant des lieux denfouissement afin

dvaluer la contribution de ces missions au rchauffement de la Terre et de dvelopper des techniques optimales de

contrle des missions provenant des lieux denfouissement individuels. La mthode la plus populaire disponible pour

les points de mesure des gaz de lieux denfouissement, la technique de la chambre de flux dbit statique ou ferm,

prsente des imprcisions associes la technique elle-mme. Une nouvelle approche de modlisation mathmatique est

prsente afin destimer les erreurs causes par les changements au cheminement du dbit des gaz dans la couverture

de sol lorsquune chambre de flux dbit ferm est place sur le dessus de cette couverture pour mesurer les missions

gazeuses. Le modle numrique deux dimensions propos considre les procds dadvection et de dispersion de la

migration des gaz. Le modle a t talonn et vrifi en utilisant des donnes provenant dexpriences contrles en

laboratoire. Il peut prdire les dbits avec une prcision de 95 %. Notre analyse, base sur des expriences en laboratoire

et des prdictions de modles, montre que la technique par chambre dbit ferm sous-estimerait le dbit rel de gaz par

autant que 65 %, selon la configuration de la chambre dbit, le dbit rel ainsi que les caractristiques du sol dans la

couverture.

Mots cls : gaz denfouissement, mthane, rchauffement de la Terre, modlisation mathmatique, couche de couverture,

simulation, surveillance environnementale, chambre de flux.

[Traduit par la Rdaction]

Introduction

Sanitary landfills accepting biodegradable organic waste are

known to produce significant quantities of methane (CH4 ) and

carbon dioxide (CO2 ). Although both CH4 and CO2 are green-

basis because it has a global warming potential (GWP) of 23

with reference to a 100-year time horizon (IPCC 2001), as compared to CO2 with a GWP of 1. According to some estimates,

sanitary landfills contribute up to 12% of global anthropogenic

Received 11 October 2001. Revision accepted 7 October 2002. Published on the NRC Research Press Web site at http://jees.nrc.ca/ on

4 December 2002.

M.D.N. Perera. Stantec Consulting Ltd., 1122 - 4th Street S.W., Calgary, AB T2R 1M1, Canada.

J.P.A. Hettiaratchi1 and G. Achari. Department of Civil Engineering, The University of Calgary, 2500 University Drive N.W., Calgary,

AB T2N 1N4, Canada.

Written discussion of this article is welcomed and will be received by the Editors until 31 March 2003.

1

DOI: 10.1139/S02-033

452

CH4 emissions into the atmosphere (Bogner et al. 1997). Considering the importance of controlling atmospheric CH4 emissions from landfills, there is renewed interest in quantifying

surface emissions from individual landfills. Accurate quantification of surface gas emissions is required for three purposes:

(i) to assess the magnitude of the problem created by individual landfills and the global problem posed by land disposal

of biodegradable organic waste, (ii) to identify and implement

methods to control atmospheric emissions, and (iii) to evaluate

the effectiveness of methods to control atmospheric emissions

of landfill gases.

Although several techniques are available for measuring surface gas emissions from landfills, the closed, or static, flux

chamber method is the most popular (Bogner et al. 1997; Czepiel

et al. 1996; Livingston and Hutchinson 1995). A closed flux

chamber is used to measure the landfill gas flux at a selected

point on the surface of a landfill. The flux is measured at the

point of emission into the atmosphere, after the gas generated

within the landfill had the opportunity to travel across the landfill cover layer. Although this method is simple and applicable

in almost all landfill situations, various researchers have identified errors in surface flux calculations based on closed flux

chamber measurements (Reinhart et al. 1992).

The closed flux chamber technique involves trapping the gas

in an enclosed chamber as it leaves the soil surface and allowing

the gas concentration to build up within this chamber. The rate of

gas emission into the chamber, F in mol m2 s1 , is calculated

using

V dC

[1]

F =

A dt

where V (m3 ) and A (m2 ) are volume and area of the flux chamber, respectively, and C (mol m3 ) is the concentration of the

component gas inside the chamber.

Since gas collected inside the closed flux chamber is not being

continually diluted with external air (as is the case with the open

flux chamber technique), even small fluxes can be measured

with this method. However, the main drawback of the closed

chamber technique is the influence of the chamber itself on the

flux rate. As gas accumulates in the chamber, the increase in

concentration decreases the flux rate. Therefore, this technique

tends to underestimate the gas fluxes. According to Matthias et

al. (1978), the actual flux could be as much as 55% higher than

the flux calculated from flux chamber measurements.

Several researchers have proposed error correction methods

to improve the calculated surface flux using closed flux chamber measurements. Hutchinson and Mosier (1981) proposed

the use of three chamber measurements as a correction. Since

their correction considers only diffusive gas flow, it has limited

use in a landfill situation where flow may be controlled by advection, mechanical dispersion, and molecular diffusion. Other

researchers (e.g., Matthias et al. 1978; Hutchinson and Mosier

1981; Healy et al. 1996) have attempted to use mathematical

models to improve the accuracy of gas flux determinations.

Matthias et al. (1978) and Hutchinson and Mosier (1981) proposed the use of separate nonlinear models, each based on the

theory of molecular diffusion in soils. The model proposed by

Matthias et al. (1978) is capable of simulating nitrous oxide

(N2 O) flux from homogeneous soil in the presence of a closed

flux chamber. It is a two-dimensional diffusion model under the

Cartesian coordinate system, but it does not consider advective

or mechanical dispersion induced flow of gas in soils. The most

extensive numerical modeling effort was by Healy et al. (1996).

They considered three-dimensional molecular diffusion but also

neglected advective flow as well as flow due to mechanical dispersion. Their model is applicable only to diffusion of a single

gas in a porous medium under ambient pressure.

Since most of the past modeling efforts are directed at low

fluxes of N2 O from soils by diffusive flow alone, the resulting models cannot be applied in landfill situations where high

fluxes could be encountered. Pressures in landfills could be

much higher than atmospheric pressure, and therefore both advective and dispersive (i.e., molecular diffusion and mechanical

dispersion) processes govern gas flow behavior across a landfill cover. There have been instances where the pressure inside

landfills exceeded 2 kPa (Hartless 1995). Furthermore, landfill

gas is a mixture of gases, including CH4 and CO2 as well as a

variety of other gases in trace quantities.

In this paper, a mathematical modeling approach is introduced that can be used to estimate the errors associated with

calculating landfill surface flux from closed flux chamber measurements. The approach requires the collection of data using

the closed flux chamber and applying a new two-dimensional

advectivedispersive gas migration model. The mathematical

model formulation is presented first. The two-dimensional gas

migration model was solved numerically using the Alternating

Implicit Direction (ADI) method. Model calibration, sensitivity analysis, and verification using laboratory data sets are then

presented. The paper concludes with a description of potential

model applications, strengths, and limitations.

Theoretical formulations

The errors associated with the determination of surface gas

fluxes using flux chambers can be quantified by modeling gas

migration in soil in the vicinity of the flux chamber considering gas flow across a landfill cover to be multicomponent and

unsteady. The theoretical aspects of the mathematical model

that accommodates the two physical processes of importance,

advection and dispersion, are discussed below.

Landfill cover systems are usually layered structures, with

different layers serving different purposes. In this analysis, for

simplicity, it is assumed that a landfill cover consists of a single

layer of soil. Gas migration in the soil cover is modeled assuming soil is a continuum. Since flux chambers considered in this

study are cylindrical, perturbations in the soil atmosphere are

radial. However, it can further be simplified to an axisymmetric

case assuming the soil properties are isotropic on the horizontal

plane.

2002 NRC Canada

Perera et al.

453

Landfill Cover

the soil (Note: soil respiration that could be a source of CO2 is

neglected). Since the intent of this paper is to assess the effects

of only the physical processes (i.e., advection and diffusion) on

flux chamber measurements, the sourcesinkreaction term S

will not be used in subsequent equations. In addition, the model

development has been further simplified by using only two gas

components, CO2 and air, with the air component representing CH4 , N2 , O2, and other trace gases. The reason behind this

simplification is that a single nonreactive gas (i.e., CO2 ) is sufficient to assess the effects of advective and dispersive processes

on flux chamber measurement, and these processes are the only

ones contributing to errors in gas measurement.

Gas velocities are determined from Darcys law

[3]

Waste Layer

of a closed flux chamber. In this analysis, only gas migration is

considered and the moisture migration is neglected. Moisture

migration is neglected for a number of reasons. These include

the short time frame of flux chamber measurement (typically,

flux chamber measurements are taken quickly to minimize perturbations in the soil atmosphere) and the relatively low moisture flux because of low hydraulic head and hydraulic conductivity of landfill cover soils. However, the effect of pore water

on gas permeability is taken into account during mechanistic

modeling. At a landfill cover, the gas phase may include CH4 ,

CO2 , oxygen (O2 ), nitrogen (N2 ), and water vapour. In model

development, the potential dissolution of gases in pore water

and the presence of water vapour in gas phase are neglected.

Advective flows due to gravity and differential density are also

neglected because of the small vertical distances considered in

modeling.

The axisymmetric relationship to describe the fate and transport of a fluid in a uniform soil is given in eq. [2] (Mendoza and

Frind 1990). When applied to gas movement in the presence

of a flux chamber, the z direction is vertically downwards and

the r direction is radially outwards from the centre of the flux

chamber.

Cm

(ur Cm ) (uz Cm ) Dr

Cm

[2]

+

r

t

r

z

r r

r

+ Dz

2 Cm

S

z2

krel kj P

j

where krel is the relative permeability; kj is the intrinsic permeability in j direction (m2 ); is the gas mixture viscosity (Pa s);

P is the pressure (Pa); and j = r, z.

The gas mixture viscosity () is calculated using viscosities

of the individual gases (m ) and their mole fraction (Ym ).

[4]

2

m

2

m=1 1 +

ml YYml

l=1

m=l

in which

0.5

ml = 8

Mm 0.5

1+

Ml

2

m 0.5

Ml 0.25

1+

l

Mm

Mm and Ml = molecular weight of gas m, and l, respectively.

Gas migration modeling is complex because of the interdependence of flow and transport equations. The solution procedure is therefore iterative for any time step (Mendoza and Frind

1990). In the process, the flow and transport equations are combined by deriving the pressures from concentrations of the component gases using the equation of state for gases (Findikakis

and Leckie 1979). It results in a single nonlinear equation with

concentrations of different gases as the variables.

From the equation of state for gas

[5]

P = RT

2

Cm

m=1

m3 );

m = CH4 ,

CO2 , O2 , N2 (all these gases should be considered when simulating landfill gas emissions); Dj is the coefficient of dispersion

in j direction (m2 s1 ); uj is the gas velocity in j direction (m

s1 ); j = r, z; is the gas porosity; S is the source, sink, or

reaction (decay or production rate) (mol s1 m3 ).

uj =

is the absolute temperature (K), and m represents the two gas

components, CO2 and air.

[6]

D = D + |u|

2002 NRC Canada

454

where D is the dispersion coefficient (m2 s1 ), D* is the molecular diffusion (m2 s1 ), is the dispersivity (m), and u is the

gas velocity (m s1 ).

The complete equation for a component gas i is

[7]

Cm

krel kr P Cm

krel kr 2 P

=

+

Cm

t

r r

r 2

krel kz P Cm

krel kz 2 P

+

+

Cm

z z

z2

2 Cm

2 Cm

Dr Cm

+

D

+ Dr

+

z

r 2

r r

z2

(CO2 and air). In the steady state case, the left-hand side of

eq. [7] becomes zero, and on the right-hand side all terms with

r are dropped because all of the variations are in the vertical

z direction. The steady state solution is used to determine the

initial conditions for the transient (axisymmetric) case.

[8]

n+1

Ci,j

n

Ci,j

t

n+1

n+1

Ci+1,j

Ci1,j

Boundary conditions

The boundary conditions are illustrated in Fig. 3. There are

N + 1 nodes in the r direction and M + 2 nodes in the z direction.

The number of grid points inside the chamber is defined as M1.

[9]

The governing equations were solved using the finite difference method. For higher accuracy, space derivatives of concentrations and pressures were determined by central difference

method. The nonlinear problem was linearized prior to solving

by lagging the coefficients. The linearization was achieved by

taking the pressures and viscosities from the earlier time step.

For the axisymmetric case, the Alternating Implicit Direction

(ADI) method was used (Aziz and Settari 1979).

The finite difference equation for a particular gas component

is given in eq. [8]. In this equation, n and n + 1 represented the

time step of that particular variable.

Variable grid spacings were used to reduce the number of

grid points without compromising representation of the actual

conditions. This is because a thin layer of grid points is required

to represent gas transport adjacent to the atmosphere and flux

chamber, and a thin column of grid points is required to simulate

transport in soil adjacent to the chamber wall. Figure 2 shows

the finite difference grid system simulated in the model. The

entire depth of the landfill cover was modeled.

n

n

Pi,j

Pi+1,j

n Pn

Pi,j

i1,j

krel kr

ri1 + ri

ri1 + ri

ri

ri1

n+1

n+1

n

n

Ci,j

Pi,j

Pi,j

Ci,j

+1

+1

1

1

2

k

k

rel z

C n+1 +

ri1 + ri i,j

zj 1 + zj

zj 1 + zj

n

n

n

n

2

krel kz Pi,j +1 Pi,j Pi,j Pi,j 1

+

C n+1

zj

zj 1

zj 1 + zj i,j

n+1

n+1

n

n

n Cn

Ci+1,j

Ci,j

Ci,j

C

C

i+1,j

i1,j

i1,j

2

Dr

+ Dr

+

ri

ri1

ri1 + ri

r

ri1 + ri

n

n

n Cn

Ci,j

Ci,j

+1 Ci,j

i,j 1

2

+ Dz

zj

zj 1

zj 1 + zj

krel kr

n

n

Pi1,j

Pi+1,j

Numerical solution

to both transient and steady state cases. The CO2 and air concentrations were determined considering the atmospheric gas

composition of 0.03% CO2 , 21% O2 , 78% N2 , and 1% Ar.

Top boundary: inside the chamber

Since gases accumulate with time, the gas concentrations

inside the chamber are transient. Therefore, this boundary condition is a time dependent one. The boundary condition was

determined by considering advective and diffusive flows into

the chamber until a particular time is elapsed. The flow was

2002 NRC Canada

Perera et al.

455

Bottom boundary

Flux

Chamber

Centre Line

Finite

Difference

Grid

r-Direction

the top of the waste layer. Therefore, the bottom boundary condition should characterize gas emissions from the waste layer.

To incorporate this situation, the bottom boundary condition is

given as a gas flux. The advective and diffusive fluxes at the

bottom were set to constant values.

z-Direction

[14]

Landfill

Cover

n+1

DN

v N CN

C

= QBottom

z

Therefore,

Waste

qm dAdt

[10]

Cm,z=0,t=t = Cm,atm +

R 2 H

where

qm dA dt is the volume of component m entering

the chamber during time t (mol m2 s1 ), R is the radius of the

chamber (m), and H is the height of the chamber (m).

The finite difference form of this equation can be written as

[11]

n+1

n

Ci,1

= Ci,1

+

n Cn

Ci,2

i,1

n+1

D

+ v1,1/2 Ci,1

z1

(ri + ri+1 )

2ri t

/ R 2 H

2

Left and right boundary conditions

The central axis of the chamber is characterized by the left

boundary of the domain. Due to symmetry, the concentration

gradient at this point is zero.

C

[12]

=0

r r=0

Therefore, C1,j = C3,j at any time where j = 1, N + 1.

The right boundary is selected such that there is no perturbation due to the chamber at this boundary point. Therefore, this

boundary was defined as a zero concentration gradient.

C

[13]

=0

r r=L

where L is the distance to the right boundary from the center

of the chamber (m).

Therefore, CM+1,j = CM,j at any time where j = 1, N + 1.

[15]

n+1

CN+1

=

DN

vN (zN + zN1 ) n+1

n+1

+ CN1

+

CN

DN

In these eqs. [14] and [15], QBottom is the gas flux rate at the

bottom boundary. The bottom boundary in the finite difference

grid is represented by the Nth row. QBottom is a negative value

(mol m2 s1 ) for CO2 and 0.0 for air (in a landfill setting,

fluxes for CH4 and CO2 were specified and fluxes for O2 and

N2 are set to zero).

Initial conditions

For the transient case, initial conditions are required. Since

the initial condition is the concentration at steady state, the required concentrations were determined from steady state modeling. Assuming there was no spatial heterogeneity, the steady

state was modeled as a one-dimensional situation. The top and

the bottom boundaries were similar to the two-dimensional case

outside the chamber. The governing equation for steady state

gas transport was derived from eq. [2]. It is the one-dimensional

gas migration equation in the z direction.

Removing the terms in the radial direction, the following

equation was obtained:

[16]

0=

Cm

vz Cm D

+S

z

z

solved.

Finite difference equations for steady state were derived from

a transient one-dimensional model when there was negligible

difference between the concentrations in two consecutive time

steps. For high accuracy, the Crank-Nicholson scheme was used

in the steady state solution.

The steady state equation is given in eq. [17]

2002 NRC Canada

456

[17]

vj +1/ 2

vj 1/ 2

Dj +1/ 2

Dj 1/ 2

+

Cjn 1

2(zj + zj +1 ) 2(zj + zj +1 ) zj (zj + zj +1 ) zj 1 (zj + zj +1 ) t

vj +1/ 2

vj 1/ 2

Dj +1/ 2

Dj 1/ 2

+

=

+

Cjn+1

+

Cjn1

+1

2(zj + zj +1 ) zj (zj + zj +1 )

2(zj + zj +1 ) zj 1 (zj + zj +1 )

methods

For model calibration and verification, known landfill gas

fluxes were needed. Such data could not be obtained from actual

landfills. Field experiments that simulated actual landfill conditions are difficult to control. Because of these difficulties, laboratory experiments were undertaken to generate experimental

data for model calibration and verification. The laboratory experimental data were generated using three gas flux rates and

with three flux chambers of different configurations. Since our

interest was on evaluating advective and dispersive gas migration patterns on the flux chamber measurements (as mentioned

in the model development section), it was sufficient to use one

type of gas that is nonreactive (i.e., CO2 ) to generate laboratory

data.

water (04 kPa) with an accuracy of 3% full scale. It measured

low gauge pressure using a phosphor bronze diaphragm.

Three 3/8-in. nylon tubes were installed vertically within the

soil to obtain gas samples from different depths. The tubes

reached depths of 16.5, 36, and 54 cm (see Fig. 4). The tubes

were perforated at the bottom (up to a length of 38 mm), and

a fine steel mesh was wrapped as a screen around the perforated tube to prevent entry of fine soil particles. Swagelok fittings were attached to the top of these tubes to measure the gas

concentrations. Three probes were also installed at three levels in the soil to monitor moisture content and the temperature.

Pressure inside the chamber was measured using an inclined

micromanometer.

Data collection

Experimental setup

Laboratory experiments were conducted under controlled

conditions using a soil cell constructed of a 159-L steel barrel with a height of 1 m (see Fig. 4). The bottom of the cell was

filled with a gravel layer overlying a perforated pipe bent into a

circle of diameter 25 cm. The gravel layer and a fine mesh (sieve

size 30) support the soil above and distribute the gas uniformly

within the soil cell. A port was attached to the bottom to supply

gas. This experimental setup was thoroughly checked for any

gas leaks before filling the soil. The top of the cell was kept open

to the atmosphere to enable measurement of flow rates using

a flux chamber. The soil used was a sandy loam (according to

USDA classification) with 63% sand, 22% silt, and 6% clay

.The soil was placed in three layers and compacted following

the modified Proctor compaction test procedure. The difference

in pore sizes, and therefore the contrast in gas permeabilities

between soil and gravel, result in uniform gas concentrations at

the boundary. Ultrahigh purity CO2 gas (supplied by Praxair)

was used as feed gas.

Instrumentation

The gas flow rate at the base of the cell was regulated using

needle valves and measured with a Cole-Parmer 65 mm direct

reading flow metre with a range of 0100 mL/min (accuracy

of 5% full scale). The feed gas was sent to the bottom of the

cell via TygonTM tubing. A pressure gauge was connected to

measure the gas pressure inside the cell using a Y connector.

for the following reasons: (i) to allow the assessment of the accuracy of closed flux chamber measurements (under different

flux rates and with different configurations of flux chambers)

and (ii) to calibrate the mathematical model. Both of these objectives were accomplished using the same data sets.

For model calibration in the steady state and transient state,

two types of data were collected. In the steady state, the gas flow

was assumed to be one dimensional. The test gas was sent from

the bottom of the cell for several weeks to allow equilibrium

to take place before actual readings were taken. During experimentation, the gas concentrations were measured at various

depths within the soil cell. At each depth, gas samples (2 mL)

were collected and analyzed using a Hewlett Packard MicroGas Chromatograph with thermal conductivity detectors. The

carrier gas used was helium (He). All peaks were quantified on a

personal computer with Hewlett Packard EZ-Chrom integration

software.

Once steady state was achieved within the test cell, flux chamber measurements were taken. Three flux chambers of dimensions 25 cm diameter and 16.5 cm height, 10 cm diameter and

5 cm height, and 20 cm diameter and 12 cm height were used

to generate transient case data. The use of flux chambers with

different dimensions allowed the assessment of the effect of

chamber configuration on the flux measurement. During flux

measurements, the flux chambers were placed on the soil surface at the center of the cell to ensure axisymmetry. After a

flux chamber was placed on the soil surface, the soil around the

edge was pressed to ensure a good seal. The concentrations of

2002 NRC Canada

Perera et al.

457

Inside the

Chamber

Time Dependent

Concentration

Chamber

C = Catmospheric

j=1

Right

Boundary

Left

Boundary

C =0

r

C

=0

r

i=1

i=M+2

j=N+1

Imaginary

Grid

Points

Bottom Boundary

Const. Flux

develop concentration vs. time graphs. The CO2 concentrations

inside the flux chamber were measured using an EGM2-CO2

analyzer. This instrument used a nondispersive infrared analyzer to measure CO2 concentration. The measurement range

was 050 000 ppm by volume. Within this range, the precision

of the instrument was 0.5% of full scale.

assumption because dry CO2 gas was sent through the soil for

several weeks before readings were taken (the dry gas removes

moisture from the soil and makes it almost completely dry).

The total porosity ( t ) was determined using the estimated dry

density of the soil at the time of compaction. The following

equation (American Society of Agronomy 1982) was used for

this purpose:

Model calibration

[18]

The results from laboratory experiments conducted at different gas flow rates were used for model calibration. The data

generated consisted of gas concentrations at different depths of

the soil cell and gas concentrations inside the flux chamber at

different time intervals.

During model calibration, selected model input parameters

were adjusted until there was agreement between the model

output results and the experimentally determined values.

Model calibration for steady state

The parameters used for model calibration for steady state

included soil porosity, soil diffusion coefficient, vertical and

horizontal dispersivities, and vertical and horizontal permeabilities. The model outputs included CO2 concentrations at various

locations within the soil cell. At the beginning of the calibration, the best estimates of input parameters were used. Some of

the parameter values were determined from laboratory analysis,

whereas the others were obtained from the literature.

The soil porosity was determined from the total porosity

and assuming almost completely dry soil conditions (with only

34% volumetric water content). It was possible to make this

t = 1

d

s

density (kg m3 ).

Assuming a particle density of 2650 kg m3 , and using the

measured dry density of 1634 kg m3 , the total porosity was

found to be 0.383. Therefore, the gas porosity was determined

to be 0.35 (assuming a volumetric water content of 34%).

The gas diffusion coefficient values available in the literature

are primarily for gas diffusion in air (Reid et al. 1987) but not

for porous media. To determine the coefficient of molecular

diffusion in soil, an empirical equation proposed by Millington

and Quirk (1961) was used

10/3

[19]

Dsoil = Dair

a

t2

Dair is the molecular diffusion coefficient in air (m2 s1 ), a

is the gas porosity, and t is the total porosity. Using a binary

diffusion coefficient for CO2 in air of 1.8 105 m2 s1 (Reid

et al. 1987), the soil gas diffusion coefficient was determined

to be 3.7 106 m2 s1 .

2002 NRC Canada

458

Compacted

Soil

To the Flow

Meter/Cylinder

100 cm

Nylon

Pipes

Soil Column 80 cm

Thermocouples

Fine Screen

Perforated

Ring

Pressure Gauge

Gravel

45 cm

When parameter values were changed during the model calibration process, the sensitivity of the individual parameters was

also considered. When the uncertainty of a parameter was low,

the parameter was kept unchanged. That was the case with gas

porosity and the coefficient of molecular diffusion. The other

parameters (e.g., vertical and horizontal dispersivities and vertical and horizontal permeabilities) showed high sensitivity and,

therefore, were varied within the ranges reported in the literature

until the simulated concentrations matched the experimentally

determined CO2 concentrations.

Laboratory results showed a high dispersion of gas within the

system.Although dispersion is caused by molecular diffusion as

well as mechanical dispersion, the contribution from molecular

diffusion was too small to account for this dispersion. When the

gas velocities are comparatively large, in the range of 1 m d1 ,

mechanical dispersion becomes important. Therefore, the dispersivity value was changed until the simulated results matched

the experimentally determined CO2 concentration values. The

resulting calibrated dispersivity value was 0.5 m, which lies on

the high side of the dispersivity range reported in the literature,

i.e., 0.011.0 m for the scale of this experiment (Mendoza and

Frind 1990). The lateral dispersivity was assumed to be 0.05 m.

Gas permeability did not affect the concentration profile significantly. However, it affected the flux rates due to advection

and gas pressure within the soil cell. At the bottom of the cell,

gas pressures were low (510 Pa), indicating a high intrinsic

permeability of soil. The selected intrinsic vertical permeability for calibration was 1 1010 m2 , which is equivalent to a

gas permeability of 2 104 m s1 . This value is reasonable

for sandy loam (Zheng and Bennett 1995). Horizontal permeability was assumed to be five times higher than the vertical

permeability, or 5 1010 m2 .

Parameter

Value

Intrinsic permeability horizontal

Molecular diffusion coefficient

Dispersivity vertical

Dispersivity horizontal

Volumetric air content

1 1010 m2

5 1010 m2

5.46 105 m2 s1

0.5 m

0.05 m

0.35

presented in Table 1. The resulting experimental and simulated

steady state CO2 concentration profiles along the test cell are

presented in Fig. 5. The concentration profiles were generated

for the CO2 flux rate of 199 g m2 d1 (without a flux chamber).

The model simulated results and experimental data show good

agreement.

Model calibration for transient state

The calibration for transient state was carried out in conjunction with the calibration for steady state. In the transient case,

the input parameter values were varied to obtain a reasonable

agreement between the experimentally derived CO2 concentrations within the flux chamber (as a function of time) and

the model simulated values. The set of input parameter values

presented in Table 1 provided a reasonable match.

The set of experimental data generated during the use of the

medium-sized flux chamber with a flux rate of 199 g m2 d1

was selected for the transient state model calibration. The model

simulations and experimental results are compared in Fig. 6.

2002 NRC Canada

Perera et al.

459

and model results.

35

30

30

CO 2 Concentration (%)

35

25

20

15

Experimental

10

20

15

dz = 2 cm

dz = 1 cm

10

dz = 6 cm

Simulated

25

0

0

0

0

0.2

0.4

0.1

0.2

0.6

0.3

0.4

0.5

Depth (m)

Depth (m)

Fig. 8. Sensitivity on intrinsic permeability.

500

1000

35

CO2 Concentration (%)

CO 2 Concentration (%)

chamber.

0.9

0.8

0.7

0.6

0.5

0.4

Simulated

0.3

Experimental

0.2

0.1

0.0

30

25

20

15

k = 1E-10

10

k = 1E-11

k = 1E-9

5

0

0

0.1

0.2

0.3

0.4

0.5

0.6

Depth (m)

1500

Time (s)

data (< 5% error between the two sets of results) until the CO2

concentration within the chamber reaches about 0.5%. Beyond

this gas concentration (which corresponds to a time of about

10 min), the model fails to simulate the experimental results.

Sensitivity analysis

A sensitivity analysis was undertaken to evaluate the sensitivity of model results to the following key model input parameters: grid size, intrinsic permeability (k), dispersion coefficient

(D), gas porosity (a ), and dispersivity (). During this analysis, calibrated values of all the parameters were systematically

changed (one parameter at a time) within an established plausible range. The magnitude of change in model output CO2 concentration due to a change in one of the parameters is a measure

of the sensitivity of the solution to that particular parameter. The

sensitivity analysis was undertaken for the steady state model

because the transient model took a relatively long time to run

(sometimes more than 10 h). The resultant gas concentration

profiles could be compared with either the calibrated model results or the experimental results. Since calibrated model results

showed excellent agreement with experimental results, in this

case comparisons were made with the calibrated model results.

The effect of varying model grid sizes (1, 2, and 6 cm) on gas

flux is presented in Fig. 7. There is no discernible change in CO2

concentration along the cell depth as the grid size was varied.

The low sensitivity indicates the possibility of decreasing the

number of grid points in the model, thereby reducing the model

run time. However, a larger grid spacing was not preferred for

the two-dimensional model. Since the flux chamber wall has

to be simulated in the r direction, a large grid spacing is not

appropriate. The grid spacing used for the model was 2 cm.

The sensitivity analysis result for intrinsic permeability is

shown in Fig. 8. The intrinsic permeability was varied by two

orders of magnitude, between 109 and 1011 m2 , but the model

results indicate minimal effect. The minimal significance indicates that advection could be a less important mechanism than

dispersion under the test conditions and the time interval considered during simulations.

Figure 9 illustrates the sensitivity of gas flux to the diffusive

properties by plotting the CO2 concentration profile against the

molecular diffusion coefficient, D*. The sensitivity of the model

output to D* is significant, and indicates the importance of diffusion as a process controlling gas flux rates. The sensitivity

increases with depth where the CO2 concentrations are higher.

The sensitivity of the CO2 concentrations to gas porosity is also

high (Fig. 10). Since Dsoil (a )10/3 , the high sensitivity is

expected. The parameter that exhibited the highest impact on

model outputs is dispersivity, (Fig. 11). Therefore, the sen2002 NRC Canada

460

Fig. 12. CO2 concentrations over time in the large flux chamber.

CO2 Concentration ( %)

35

30

25

20

15

D* = 3.7E-5

10

D* = 3.7E-4

D* = 3.7E-6

5

0

0

0.1

0.2

0.3

0.4

0.5

0.6

0.8

0.7

0.6

0.5

0.4

0.3

Simulated

Experimental

0.2

0.1

0.0

Depth (m)

40

1500

Fig. 13. CO2 concentrations over time in the small flux chamber.

35

1.2

30

25

20

15

phi = 0.35

10

phi = 0.45

phi = 0.25

0

0

0.1

0.2

0.3

0.4

0.5

0.6

Depth (m)

alfa = 0.65

alfa = 0.1

60

50

40

30

20

10

0

0.6

Simulated

0.4

Experimental

0.2

500

1000

1500

Time (s)

alfa = 0.35

70

0.8

alfa = 0.50

80

1.0

0.0

90

CO 2 Concentration (%)

1000

Time (s)

500

0.2

0.4

0.6

Depth (m)

mechanism under the test conditions and the time interval considered during simulations.

Model verication

For verification of the model, the calibrated model was used

to reproduce a set of experimental observations independent of

those observations used in model calibration. The experimental

results generated during the use of the large-sized flux chamber

(i.e., 25 cm diameter and 16.5 cm height) were selected for this

purpose. The simulated and experimental results are presented

and the experimental data.

For further verification of the model, a second data set obtained using the small chamber (i.e., volume 400 mL and height

5 cm) was also compared with the model results (see Fig. 13).

In this case, the model over-estimated the CO2 concentrations

within the chamber after 10 min. Thus, if the flux chamber

size is large, the model and experimental results agree up to

20 min, which is the total duration of the experiment, whereas

if the chamber size is small, the model and experimental results

agree only up to about 10 min.

limitations

Table 2 presents a comparison of flux rates calculated from

closed flux chamber measurements and the actual flux rates

used in the experimental component of the study. The percentage errors of measurement for the three types of flux chamber

configurations and the three flux rates are also presented in this

table. The results clearly show that the errors in the flux chamber measurement largely depend on the chamber size and the

flux rate. The errors associated with the measurement technique

within the flux ranges tested varied from approximately 15%

to 65% of the actual flux rate. The large flux chamber and the

smaller flux rates produce the least error in measurement. Since

2002 NRC Canada

Perera et al.

461

Table 2. Actual flux rates and measured flux rates.

Chamber

configuration

rated (mol m2 d1 )

rate (mol m2 d1 )

Error in

measurement (%)

Smalla

2.25

4.5

2.25

4.5

4.5

6.75

0.78

1.62

1.81

3.37

3.83

4.77

65.5

64.0

19.4

25.1

14.9

29.4

Mediumb

Largec

a

Chamber dimensions: 20 cm diameter and 12 cm height.

c

25 cm diameter and 16.5 cm height.

d

Measured as the gas enters the test cell (N.B. the model predicted all of these

values within a 5% error margin).

b

Fig. 14. CO2 concentration contours after 1 min and 20 min (values are in mol/L).

(a)

the model could effectively reproduce the actual flow rates, the

model is being used here to estimate the errors in flux chamber

measurements. It should be noted that the comparison of model

outputs and experimentally derived concentrations within the

flux chamber have shown that the mathematical model is capable of predicting the actual gas flux rates with a maximum error

of 5%.

In a field situation, if the model input parameters are known,

the flux chamber measurements can be corrected using the

model simulations to determine the actual flux rate of gas emissions from the surface of a landfill cover. If large-sized flux

chambers are used, the model can be used to correct the values

(b)

obtained within the first 20 min. On the other hand, with smallsized flux chambers the model can be used to increase accuracy

of measurements within the first 10 min. The experimental and

model simulation results also imply the need to use large-sized

flux chambers in the field to obtain accurate gas measurements.

An example of further model simulation is shown in Fig. 14.

This figure shows the effect of flux chamber on individual velocity vectors of gas at times 1 and 20 min. As gas pressure builds

up within the flux chamber, the flow lines of CO2 are significantly affected. The effect is small at the beginning (at time 1

min) but high after 20 min. A further strength of the model is

its potential to be used to develop sets of curves to determine

2002 NRC Canada

462

errors for any type of closed flux chamber and a range of possible flux rates. Such curves could be used to eliminate errors

(without running the model) and select the best configuration

of the flux chamber for a given range of flux rates.

At this stage of the development of the model, several limitations exist. First, the model cannot be directly applied in a

field situation where local nonhomogeneities exist in the landfill cover. Most landfill covers encountered in practice consist

of several layers of soil with variable soil properties. Consideration of flow of landfill gas through multiple layers of soil could

further enhance the predictive capabilities of the model. Second,

the application of the model requires accurate determination of

model parameters. Uncertainties in model parameter values affect the accuracy of the results. Third, inclusion of other gases,

such as CH4 , may require consideration of other processes impacting gas flow in landfill cover soils. In the case of CH4 ,

methanotrophic bacteria in landfill cover soil are known to biologically oxidize CH4 to CO2 (Stein and Hettiaratchi 2001).

This may require inclusion of a reactive term, S, in the model.

However, if advection and dispersion are the primary mechanism governing flow, biological oxidation may not provide

changes to flow patterns. Since changes to flow patterns determine the errors associated with flux chambers, consideration of

a single gas (such as CO2 ) may be sufficient to eliminate errors

associated with most flux chamber measurements.

Conclusions

The lack of sufficiently accurate and convenient surface gas

emission measurement techniques limits the capacity to estimate surface emissions of gases from landfills. The simplest and

the most popular technique for point estimation of landfill gas

fluxes, the closed flux chamber technique, provides inaccurate

results under certain situations. The closed flux chamber, once

placed on top of a landfill cover during a measurement, alters the

concentration gradient in the soil and thereby underestimates

the actual emission rate. The errors largely depend on the configuration of the flux chamber and the flux rate. The smaller the

chamber and higher the flux rate, the higher the percentage error in measurement. To minimize the errors associated with this

technique, a mathematical modeling approach is proposed. The

application of the two-dimensional advectivedispersive mathematical model will help maintain the simplicity of the flux

chamber technique, but still provide accurate estimates of CH4

emissions. The model is capable of predicting the actual flux

rates within an accuracy level of 5% and, therefore, could be

used to determine the error in a point estimate of landfill gas flux

calculated from closed flux chamber measurements. The laboratory experiments and mathematical model simulations show

that the model is sensitive to gas porosity and dispersivity, but

is fairly insensitive to grid size and intrinsic permeability.

Acknowledgements

Funding for this research was received from the Natural Sciences and Engineering Research Council (NSERC) of Canada

and the University of Calgary.

References

American Society of Agronomy. 1982. Methods of soil analysis. Part I.

Physical and mineralogical methods. 2nd ed. Edited by A. Klute.

Soil Science Society of America, Madison, Wis.

Aziz, K., and Settari, A. 1979. Petroleum reservoir simulation. Elsevier

Science Publishers, Essex, England.

Bogner, J.E., Meadows, M., and Czepiel, P. 1997. Fluxes of methane

between landfills and the atmosphere: Natural and engineering controls. Soil Use Manage. 13: 268277.

Czepiel, P.M., Mosher, B., Harriss, R.C., Shorter, J.H., McManus, J.B.,

Kolb, C.E., Allwine, E., and Lamb, B.K. 1996. Landfill methane

emissions measured by enclosure and atmospheric tracer methods.

J. Geophys. Res. 101(D11): 1671116719.

Findikakis, A.N., and Leckie, J.O. 1979. Numerical simulation of gas

flow in sanitary landfills. ASCE J. Environ. Eng. Div. 105(5): 927

945.

Hartless, R. 1995. Measuring pressures and flow of landfill gas. In

Regulation environmental impact and aftercare. Proceedings of

Sardinia 95, the 5th International Landfill Symposium, Cagliari,

Italy. Vol. III. pp. 517531.

Healy, R.W., Striegl, R.G., Russel, T.F., Hutchinson, G.P., and Livingston, G.P. 1996. Numerical evaluation of static-chamber measurements of soilatmosphere gas exchange: Identification of physical processes. Soil Sci. Soc. Am. J. 60: 740747.

Hutchinson, G.L., and Mosier, A.R. 1981. Improved soil cover method

for field measurement of nitrous oxide fluxes. Soil Sci. Soc. Am. J.

45: 311316.

IPCC (Intergovernmental Panel on Climate Change). 2001. Technical

summary. In Climate change 2001: the scientific basis. Edited by

F. Joos, A. Ramirez-Rojas, J.M.R. Stone, and J. Zillman. Cambridge

University Press, U.K.

Livingston, G.P., and Hutchinson, G.L. 1995. Enclosure-based measurement of trace gas exchange: applications and sources of error.

In Biogenic trace gases: measuring emissions from soil and water.

Edited by P.A. Matson and R.C. Harris. pp. 1451.

Matthias, A.D., Yarger, D.N., and Weinbeck, R.S. 1978. A numerical

evaluation of chamber methods for determining gas fluxes. Geophys. Res. Lett. 5(9): 765768.

Mendoza, C.A., and Frind, E.O. 1990. Advectivedispersive transport

of dense organic vapors in the unsaturated zone 1. Model development. Water Resour. Res. 26(3): 379387.

Reid, R.C., Prausnitz, J.M., and Poling, B.E. 1987. The properties of

gases and liquids. McGraw-Hill Inc., New York.

Reinhart, D.R., Cooper, D.C., and Walker, B.L. 1992. Flux chamber

design and operation for the measurement of municipal solid waste

landfill gas. J. Air & Waste Manage. Assoc. 42(8): 10671070.

Stein, V.B., and Hettiaratchi, J.P.A. 2001. Methane oxidation in three

Alberta soils: Influence of soil parameters and methane flux rates.

J. Environ. Technol. 22: 101111.

Zheng, C., and Bennett, G.D. 1995. Applied contaminant transport

modeling: theory and practice. Van Nostrand Reinhold, New York,

N.Y.

Perera et al.

List of symbols

A

C

Cm

CO2

D

Dj

D*

Dair

Dsoil

F

H

kj

krel

L

concentration of the component gas inside the chamber

(g m3 )

concentration of gas m (mol m3 )

concentration of O2 (mol m3 )

dispersion coefficient (m2 s1 )

coefficient of dispersion in j direction (m2 s1 )

molecular diffusion (m2 s1 )

diffusion coefficient in air (m2 s1 )

diffusion coefficient in soil (m2 s1 )

rate of gas emission into the chamber (g m2 s1 )

height of the chamber (m)

intrinsic permeability in j direction (m2 )

relative permeability

distance to the right boundary from the chamber centre

(m)

463

Mm

P

R

Rc

QBottom

qm

S

T

u

uj

V

X

Ym

pressure (Pa)

universal gas coefficient (8.314 J K1 mol1 )

radius of the chamber (m)

gas flux rate at the bottom boundary (mol m2 s1 )

gas flux for component m (mol m2 s1 )

source, sink, or reaction (decay, or production rate)

(mol s1 m3 )

absolute temperature (K)

gas velocity (m s1 )

gas velocity in j direction (m s1 )

volume of the flux chamber (m3 )

concentration of organisms (g m3 )

mole fraction of gas m

dispersivity (m)

gas mixture viscosity (Pa s)

porosity

Copyright of Journal of Environmental Engineering & Science is the property of Canadian Science Publishing

and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright

holder's express written permission. However, users may print, download, or email articles for individual use.

## Bien plus que des documents.

Découvrez tout ce que Scribd a à offrir, dont les livres et les livres audio des principaux éditeurs.

Annulez à tout moment.